gisfromscratch

Representation	Reflex	Model-Based	Goal-Based	Utility-Based
Atomic	✅ Month 1 – Build & test basic rule engine	🔄 Month 1–2 – Add percept memory	🔄 Month 2 – Add goal-check logic	🔄 Month 2 – Implement static utility mapping
Factored	🔄 Month 2 – Simulate feature-driven logic	🔄 Month 2–3 – Track derived variables	🔄 Month 3 – Add goal prioritization	🔄 Month 3 – Score-based decision system
Structured	🔄 Month 3 – Encode spatial logic with RDF	🔄 Month 4 – Implement model-based object tracking	🔄 Month 4 – Goal hierarchy on assets & regions	🔄 Month

gisfromscratch

Representation	Description	Example	Use Case
Atomic	Opaque percept–action pair	("thermal=True", "landcover=forest")	Basic sensors
Factored	Attribute vectors	{thermal=True, humidity=12%}	Local context evaluation
Structured	Relational models	FireEvent → Region → Assets	Spatially aware reasoning

gisfromscratch

Agent Type	Description
Simple Reflex	Reacts instantly to percepts using condition–action rules
Model-Based Reflex	Maintains an internal state to handle partial observability
Goal-Based	Chooses actions to reach specific goals based on environmental input
Utility-Based	Selects the most beneficial action based on a utility calculation

gisfromscratch

Traditional Agent	Learning Agent
Static logic	Adaptive behavior
Pre-coded policies	Feedback-driven evolution
Limited generalization	Exploratory and self-improving
High false positives	Tuned to real-world feedback

gisfromscratch

Data Source	Metric Type	Example
Satellite imagery (MODIS)	Detection accuracy	Missed or late alerts
Field reports (fire crews)	Classification validity	False alarms
Historical maps	Predictive accuracy	Over/under-estimated spread
Resource tracking systems	Efficiency	Time to containment

gisfromscratch

Component	Role
Performance Element	Makes decisions and takes actions based on current knowledge
Learning Element	Updates behavior based on feedback
Critic	Evaluates actions against objectives or ground truth
Problem Generator	Suggests new experiences for learning

gisfromscratch

Agent Type	Limitations	Utility-Based Advantage
Simple Reflex Agent	Can’t plan or weigh consequences	Evaluates future states and trade-offs
Model-Based Reflex	No decision trade-off modeling	Weighs conflicting priorities (e.g., time vs. cost)
Goal-Based Agent	All goals = equal value	Prefers better outcomes, not just goal success
Utility-Based Agent	✅ Handles uncertainty, learns over time	✅ Chooses rationally based on expected impact

gisfromscratch

Component	Wildfire Scenario Implementation
Performance	Maximize early detection, minimize false positives, reduce response time
Environment	Forests, communities, terrain, weather systems, sensor data streams
Actuators	Task UAVs, trigger alarms, reroute resources, update risk maps
Sensors	EO satellites, IR cameras, real-time wind, drone telemetry

gisfromscratch

Spatial Analysis Category	Wildfire Application
Understanding Where	Detect new hotspots from satellite feeds
Determining Relationships	Analyze wind direction relative to terrain
Finding Best Locations/Paths	Optimize drone patrol routes and safe evacuation corridors
Detecting Patterns	Spot shifting fire clusters using temporal data
Making Predictions	Model fire spread under various weather scenarios

gisfromscratch

Capability	Simple Reflex	Model-Based Reflex	Goal-Based
Fire detection	Yes	Yes	Yes
Internal state tracking	No	Yes	Yes
Future outcome simulation	No	Limited	Yes
Action planning	No	Reactive	Strategic
Goal prioritization	No	Implicit	Explicit
Resource allocation	No	Minimal	Optimized